Google connects an AI agent to maps, satellites and solar data
An experimental Earth AI workflow can combine private location lists with large geospatial datasets.
AI, technology and science
Models, tools, AI companies and research, and the chips and infrastructure behind them.
36 stories
An experimental Earth AI workflow can combine private location lists with large geospatial datasets.
Its research frameworks track characters, locations and objects across a sequence of generated shots.
Diffusion Controller changes the path from noise to a picture, with tests based on Stable Diffusion 1.4.
Your algorithm uses written interests and feedback to change recommendations, according to the company.
The three-chip family combines local vision and audio processing with a shared package design.
The system splits documents into sections, assigns review work and combines the findings into a technical critique.
The model searches more than 176,000 Latin inscriptions and gives historians parallels to examine.
Gemini Robotics On-Device runs locally and adapted to new tasks with 50 to 100 demonstrations in company tests.
Google’s research model learns acoustic patterns, but predicting sounds does not establish what they mean.
The release includes model checkpoints and evaluation code under Apache 2.0.
The API returns probabilities for fixed answers instead of generating a written response.
The four models range from 5 million to 209 million parameters and can use known future inputs.